Extracts from the leaf of the ginkgo biloba tree have been used for millennia in traditional medicine. Ginkgo may be of benefit in Alzheimer disease (AD) because of its well-studied anti-oxidant effects as well as its ability to antagonize platelet activating factor. Over the last 15 years there have been many studies evaluating its use in dementia. All the clinical studies of ginkgo in AD are European. Most of the studies did not perform what would be considered adequate clinical evaluations for the diagnosis of AD or had very limited outcome measures. Given the enormity of the societal impact of AD it is extremely important to begin to critically evaluate its use. We plan to study the effect of 2 doses of standardized ginkgo extract on cognitive function in AD. Either a negative or positive result will be important because of the widespread use of ginkgo by patients with AD and their families despite the absence of any formal American study. This pilot study will enroll 42 patients with probable AD in a randomized, double-blind placebo-controlled 32 week trial. Subjects will be placed in l of 3 groups: placebo, 120 mg per day standardized ginkgo biloba extract, or 240 mg per day of ginkgo extract. Primary outcome measures are the AD Assessment Scale -cognitive subtest and a clinician global rating scale. Secondary outcome measures include the Mini-Mental State Examination, a visual attention reaction time test, Clinical Dementia Rating Scale, activities of daily living and a behavioral scale. Data will be analyzed primarily by using change scores from baseline with the baseline measure as the covariate. In the event of a suggestive positive result, additional funding will be requested to l) extend the duration of the study; 2) "power-up" the sample size through a possible multicenter proposal; 3) further elucidate the clinical effect (which types of patients benefit); 4) further understand which component of the ginkgo extract exerts the therapeutic effect; 5) better examine secondary outcomes. As a pilot study, a negative result will obviously not have enough statistical power to prove no effect. Additional funding will be requested to increase the number of subjects in order to have confidence in a negative result.